• Korean Journal of Radiology
• /
• v.25 no.4
• /
• pp.363-373
• /
• 2024

Objective: To develop and evaluate a deep learning model for automated segmentation and detection of bone metastasis on spinal MRI. Materials and Methods: We included whole spine MRI scans of adult patients with bone metastasis: 662 MRI series from 302 patients (63.5 ± 11.5 years; male:female, 151:151) from three study centers obtained between January 2015 and August 2021 for training and internal testing (random split into 536 and 126 series, respectively) and 49 MRI series from 20 patients (65.9 ± 11.5 years; male:female, 11:9) from another center obtained between January 2018 and August 2020 for external testing. Three sagittal MRI sequences, including non-contrast T1-weighted image (T1), contrast-enhanced T1-weighted Dixon fat-only image (FO), and contrast-enhanced fat-suppressed T1-weighted image (CE), were used. Seven models trained using the 2D and 3D U-Nets were developed with different combinations (T1, FO, CE, T1 + FO, T1 + CE, FO + CE, and T1 + FO + CE). The segmentation performance was evaluated using Dice coefficient, pixel-wise recall, and pixel-wise precision. The detection performance was analyzed using per-lesion sensitivity and a free-response receiver operating characteristic curve. The performance of the model was compared with that of five radiologists using the external test set. Results: The 2D U-Net T1 + CE model exhibited superior segmentation performance in the external test compared to the other models, with a Dice coefficient of 0.699 and pixel-wise recall of 0.653. The T1 + CE model achieved per-lesion sensitivities of 0.828 (497/600) and 0.857 (150/175) for metastases in the internal and external tests, respectively. The radiologists demonstrated a mean per-lesion sensitivity of 0.746 and a mean per-lesion positive predictive value of 0.701 in the external test. Conclusion: The deep learning models proposed for automated segmentation and detection of bone metastases on spinal MRI demonstrated high diagnostic performance.

• Korean Journal of Radiology
• /
• v.23 no.10
• /
• pp.1009-1018
• /
• 2022

Objective: This study aimed to investigate the feasibility of using artificial intelligence (AI) to identify normal chest radiography (CXR) from the worklist of radiologists in a health-screening environment. Materials and Methods: This retrospective simulation study was conducted using the CXRs of 5887 adults (mean age ± standard deviation, 55.4 ± 11.8 years; male, 4329) from three health screening centers in South Korea using a commercial AI (Lunit INSIGHT CXR3, version 3.5.8.8). Three board-certified thoracic radiologists reviewed CXR images for referable thoracic abnormalities and grouped the images into those with visible referable abnormalities (identified as abnormal by at least one reader) and those with clearly visible referable abnormalities (identified as abnormal by at least two readers). With AI-based simulated exclusion of normal CXR images, the percentages of normal images sorted and abnormal images erroneously removed were analyzed. Additionally, in a random subsample of 480 patients, the ability to identify visible referable abnormalities was compared among AI-unassisted reading (i.e., all images read by human readers without AI), AI-assisted reading (i.e., all images read by human readers with AI assistance as concurrent readers), and reading with AI triage (i.e., human reading of only those rendered abnormal by AI). Results: Of 5887 CXR images, 405 (6.9%) and 227 (3.9%) contained visible and clearly visible abnormalities, respectively. With AI-based triage, 42.9% (2354/5482) of normal CXR images were removed at the cost of erroneous removal of 3.5% (14/405) and 1.8% (4/227) of CXR images with visible and clearly visible abnormalities, respectively. In the diagnostic performance study, AI triage removed 41.6% (188/452) of normal images from the worklist without missing visible abnormalities and increased the specificity for some readers without decreasing sensitivity. Conclusion: This study suggests the feasibility of sorting and removing normal CXRs using AI with a tailored cut-off to increase efficiency and reduce the workload of radiologists.

• Korean Journal of Radiology
• /
• v.24 no.9
• /
• pp.827-837
• /
• 2023

Objective: To investigate the predictive value of radiomics features based on cardiac magnetic resonance (CMR) cine images for left ventricular adverse remodeling (LVAR) after acute ST-segment elevation myocardial infarction (STEMI). Materials and Methods: We conducted a retrospective, single-center, cohort study involving 244 patients (random-split into 170 and 74 for training and testing, respectively) having an acute STEMI (88.5% males, 57.0 ± 10.3 years of age) who underwent CMR examination at one week and six months after percutaneous coronary intervention. LVAR was defined as a 20% increase in left ventricular end-diastolic volume 6 months after acute STEMI. Radiomics features were extracted from the oneweek CMR cine images using the least absolute shrinkage and selection operator regression (LASSO) analysis. The predictive performance of the selected features was evaluated using receiver operating characteristic curve analysis and the area under the curve (AUC). Results: Nine radiomics features with non-zero coefficients were included in the LASSO regression of the radiomics score (RAD score). Infarct size (odds ratio [OR]: 1.04 (1.00-1.07); P = 0.031) and RAD score (OR: 3.43 (2.34-5.28); P < 0.001) were independent predictors of LVAR. The RAD score predicted LVAR, with an AUC (95% confidence interval [CI]) of 0.82 (0.75-0.89) in the training set and 0.75 (0.62-0.89) in the testing set. Combining the RAD score with infarct size yielded favorable performance in predicting LVAR, with an AUC of 0.84 (0.72-0.95). Moreover, the addition of the RAD score to the left ventricular ejection fraction (LVEF) significantly increased the AUC from 0.68 (0.52-0.84) to 0.82 (0.70-0.93) (P = 0.018), which was also comparable to the prediction provided by the combined microvascular obstruction, infarct size, and LVEF with an AUC of 0.79 (0.65-0.94) (P = 0.727). Conclusion: Radiomics analysis using non-contrast cine CMR can predict LVAR after STEMI independently and incrementally to LVEF and may provide an alternative to traditional CMR parameters.

• Korean Journal of Radiology
• /
• v.23 no.3
• /
• pp.322-332
• /
• 2022

Objective: CT plays a central role in determining the resectability of pancreatic cancer, which directs the use of neoadjuvant therapy. This study aimed to assess the diagnostic accuracy of CT in predicting circumferential resection margin (CRM) involvement in patients with resectable or borderline resectable pancreatic head cancer. Materials and Methods: Seventy-seven patients who were scheduled for upfront surgery for resectable or borderline resectable pancreatic head cancer were prospectively enrolled, and 75 patients (38 male and 37 female; mean age ± standard deviation, 68 ± 11 years) were finally analyzed. The CRM status was evaluated separately for the superior mesenteric artery (SMA) and posterior and superior mesenteric vein/portal vein (SMV/PV) margins. Three independent radiologists reviewed the preoperative CT images and evaluated the resection margin status. The reference standard for CRM status was pathologic examination of pancreaticoduodenectomy specimens in an axial plane perpendicular to the axis of the second portion of the duodenum. The diagnostic accuracy of CT was assessed for overall CRM involvement, defined as involvement of the SMA or posterior margins (per-patient analysis), and involvement of each of the three resection margins (per-margin analysis). The data were pooled using a crossed random effects model. Results: Forty patients had pathologically confirmed overall CRM involvement in pancreatic cancer, while CRM involvement was not seen in 35 patients. For overall CRM involvement, the pooled sensitivity and specificity were 15% (95% confidence interval: 7%-49%) and 99% (96%-100%), respectively. For each of the resection margins, the pooled sensitivity and specificity were 14% (9%-54%) and 99% (38%-100%) for the SMA margin, 12% (8%-46%) and 99% (97%-100%) for the posterior margin; and 37% (29%-53%) and 96% (31%-100%) for the SMV/PV margin, respectively. Conclusion: CT showed very high specificity but low sensitivity in predicting pathological CRM involvement in pancreatic cancer.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.27 no.1
• /
• pp.116-125
• /
• 2024

The COVID-19 pandemic is not over despite the emergency use authorization as can see recent COVID-19 daily confirmed cases. The viruses are not only difficult to diagnose and treat due to random mutations, but also pose threat human being because they have the potential to be exploited as biochemical weapons by genetic manipulation. Therefore, it is inevitable to the rapid antibody-based therapeutic platform to quickly respond to future pandemics by new/re-emerging viruses. Although numerous researches have been conducted for the fast development of antibody-based therapeutics, it is sometimes hard to respond rapidly to new viruses because of complicated expression or purification processes for antibody production. In this study, a novel rapid antibody-based therapeutic platform using single B cell sorting method and mRNA-antibody. High immunogenicity was caused to produce antibodies in vivo through mRNA-antigen inoculation. Subsequently, antigen-specific antibody candidates were selected and obtained using isolation of B cells containing antibody at the single cell level. Using the antibody-based therapeutic platform system in this study, it was confirmed that novel antigen-specific antibodies could be obtained in about 40 days, and suggested that the possibility of rapid response to new variant viruses.

• Knowledge Management Research
• /
• v.25 no.1
• /
• pp.155-172
• /
• 2024

This study investigates the effectiveness of foreign language learning through diverse treatments in virtual settings, particularly by differentiating virtual environments with three textual enhancement techniques. A 2 × 3 mixed-factorial design was used, treating virtual environments as within-subject factors and textual enhancement techniques as between-subject factors. Participants experienced two videos, each in different virtual learning environments with one of the random textual enhancement techniques. The results showed that the interaction between different virtual environments and textual enhancement techniques had a statistically significant impact on presence among groups. In examining main effects of virtual environments, significant differences were observed in flow and attitude toward pre-post learning. Also, main effects of textual enhancements notably influenced flow, intention to use, learning satisfaction, and learning confidence. This study highlights the potential of Metaverse in foreign language learning, suggesting that learner experiences and effects vary with different virtual environments.

• Smart Media Journal
• /
• v.13 no.4
• /
• pp.33-48
• /
• 2024

With the advent of 5G, characterized by Enhanced Mobile Broadband (eMBB), Ultra-Reliable Low Latency Communications (URLLC), and Massive Machine Type Communications (mMTC), efficient network management and service provision are becoming increasingly critical. This paper proposes a novel approach to address key challenges of 5G networks, namely ultra-high speed, ultra-low latency, and ultra-reliability, while dynamically optimizing network slicing and resource allocation using machine learning (ML) and deep learning (DL) techniques. The proposed methodology utilizes prediction models for network traffic and resource allocation, and employs Federated Learning (FL) techniques to simultaneously optimize network bandwidth, latency, and enhance privacy and security. Specifically, this paper extensively covers the implementation methods of various algorithms and models such as Random Forest and LSTM, thereby presenting methodologies for the automation and intelligence of 5G network operations. Finally, the performance enhancement effects achievable by applying ML and DL to 5G networks are validated through performance evaluation and analysis, and solutions for network slicing and resource management optimization are proposed for various industrial applications.

• Journal of Animal Science and Technology
• /
• v.66 no.2
• /
• pp.310-325
• /
• 2024

Glyconutrients help in the body's cell communication. Glyconutrients and synbiotics are promising options for improving immune function. Therefore, we hypothesized that combining synbiotics and glyconutrients will enhance pig nutrient utilization. 150 pigs (Landrace × Yorkshire × Duroc), initially weighing 58.85 ± 3.30 kg of live body weight (BW) were utilized to determine the effects of synbiotics-glyconutrients (SGN) on the pigs' performance, feed efficiency, gas emission, pork traits, and composition of fatty acids. The pigs were matched by BW and sex and chosen at random to 1 of 3 diet treatments: control = Basal diet; TRT1 = Basal diet + SGN 0.15%; TRT2 = Basal diet + SGN 0.30%%. The trials were conducted in two phases (weeks 1-5 and weeks 5-10). The average daily gain was increased in pigs fed a basal diet with SGN (p = 0.036) in weeks 5-10. However, the apparent total tract digestibility of dry matter, nitrogen, and gross energy did not differ among the treatments (p > 0.05). Dietary treatments had no effect on NH₃, H₂S, methyl mercaptans, acetic acids, and CO₂ emissions (p > 0.05). Improvement in drip loss on day 7 (p = 0.053) and tendency in the cooking loss were observed (p = 0.070) in a group fed basal diets and SGN at 0.30% inclusion level. The group supplemented with 0.30% of SGN had higher levels of palmitoleic acid (C16:1), margaric acid (C17:0), omega-3 fatty acid, omega-6 fatty acid, and ω-6: ω-3 ratio (p = 0.034, 0.020, 0.025, 0.007, and 0.003, respectively) in the fat of finishing pigs. Furthermore, group supplemented with 0.30% of SGN improved margaric acid (C17:0), linoleic acid (C18:2n6c), arachidic acid (C20:0), omega 6 fatty acid, omega-6 to omega-3 ratio, unsaturated fatty acid, and monounsaturated fatty acid (p = 0.037, 0.05, 0.0142, 0.036, 0.033, 0.020, and 0.045, respectively) in the lean tissues of finishing pigs compared to pigs fed with the control diets. In conclusion, the combination of probiotics, prebiotics, and glyconutrients led to higher average daily gain, improved the quality of pork, and more favorable fatty acid composition. Therefore, these results contributed to a better understanding of the potential of SGN combinations as a feed additive for pigs.

• Journal of the Korean Society for Library and Information Science
• /
• v.58 no.2
• /
• pp.251-268
• /
• 2024

Recently, with the implementation of the "e-Government Performance Management Guidelines," there is a growing demand for setting performance indicators for information systems. For systems that provide information services to the public, such as CODIL, it is not easy to set performance indicators. This study presented a research model that applies Monte Carlo simulation to set expected performance targets that can be achieved through CODIL based on objective evidence. Among the survey contents conducted from 2015 to 2023, the statistical characteristics of user satisfaction regarding the frequency of use of construction technology information provided by CODIL were designated as input variables. Future expected targets and confidence intervals from 2024 to 2026 were designated as outcome variables. The expected target value was measured by generating 5 simulation alternatives and 1,000 random numbers for each alternative. Next, the measured expected goals were interpreted and compared with the results of time series regression analysis measured in previous studies. Although, as in previous studies, the expected target value could not be predicted based on time series regression analysis that considers the correlation between years. However, compared to previous studies, this study can be considered a more accurate analysis result because it predicted the expected target value based on 5,000 input variables.

• Journal of Cardiovascular Imaging
• /
• v.30 no.3
• /
• pp.153-168
• /
• 2022

Recent studies have shown that sodium/glucose cotransporter 2 (SGLT2) inhibitors might exert favourable changes on cardiac parameters as observed on cardiovascular imaging. We conducted a systematic review and meta-analysis to determine the effects of SGLT2 inhibitors on cardiac imaging parameters. Four electronic databases (PubMed, Embase, Cochrane, Scopus) were searched for studies in which the effects of SGLT2 inhibitors on cardiac imaging parameters were examined. Studies in which a population was administered SGLT2 inhibitors and analysed by echocardiography and/or cardiac magnetic resonance (CMR) imaging were included. Random-effects pair-wise meta-analysis models were utilized to summarize the studies. A total of 11 randomized controlled trials was included with a combined cohort of 910 patients. Comparing patients receiving SGLT2 inhibitors with subjects receiving placebo, the mean change in CMR-measured left ventricular mass (LVM) was -3.87 g (95% confidence interval [CI], -7.77 to 0.04), that in left ventricular end-systolic volume (LVESV) was -5.96 mL (95% CI, -10.52 to -1.41) for combined LVESV outcomes, that in left atrial volume index (LAVi) was -1.78 mL/m² (95% CI, -3.01 to -0.55) for combined LAVi outcomes, and that in echocardiography-measured E/e' was -0.73 (95% CI, -1.43 to -0.03). Between-group differences were not observed in LVM and LVESV after indexation. The only between-group difference that persisted was for LAVi. Treatment with SGLT2 inhibitors resulted in reduction in LAVi and E/e' on imaging, indicating they might have an effect on outcomes associated with LV diastolic function.